复合粒子群算法及其在动态环境中的应用  

Composite particle swarm optimization algorithm and its application in dynamic environments

作　　者：刘黎黎[1] 汪定伟[1] 

机构地区：[1]东北大学信息科学与工程学院,辽宁沈阳110004

出　　处：《系统工程学报》2011年第2期269-274,共6页Journal of Systems Engineering

基　　金：国家自然科学基金重点项目(70931001;70771021;70721001;61004121);国家自然科学基金创新群体项目(60821-063);国家教育部博士点基金(200801450008)

摘　　要：针对粒子群算法在求解动态优化问题时存在多样性缺失和搜索效率低的问题,模拟物理学中复合粒子的构成机理,提出了复合粒子群优化算法.通过对复合粒子进行构造、自调整、整体迁移等操作,使粒子充分结合其它粒子在进化中有价值的信息,并提出异速度映射机制,使粒子以较快的速度跟踪最优点的变化轨迹,保持群体多样性的同时兼顾了寻优速度.算法应用于两种动态测试函数中,取得了满意的结果.Particle swarm optimization （ PSO ） algorithm is inclined to fall into diversity loss and low searching efficiency in dynamic environments. In this paper, the features of composite particles in physics were used. Some operations such as construction, self-adjustment and integral-movement are designed for the composite particle. This proposed technique is aimed at making particles integrate valuable information from other ones sufficiently, and a new scheme called velocity-anisotropic reflection is proposed to drive particles to track the trajectory of the varying optimum rapidly, hence, this method maintains diversity as well as guarantees searching precision.The algorithm has been applied to two benchmark dynamic function problems, the satisfactory results have been achieved.

关 键 词：粒子群优化 复合粒子 异速度映射 动态环境 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]